Trees in Viz: Chrome's Centralized Compositing Engine
from Blink286 Hosted by Free Debreuil
The source details a significant re-architecture of the Chromium rendering pipeline, known as JellyMander, centered on the feature Trees in Viz (TiV). This change shifts the core task of frame compositing—the process of calculating layer positions and generating draw commands—from individual renderer processes to the centralized Viz service running in the GPU process. Renderers now serialize and transmit raw visual data, including their layer tree and associated property trees, which Viz uses to compute the final display frame. This centralization is designed to reduce CPU overhead in the content process, facilitate unified scheduling, and enable major performance optimizations, particularly for rendering across site-isolated frames. Although Phase 1 focused on remote compositing, future phases aim to fully relocate complex dynamic updates, such as smooth scrolling and compositor animations, into Viz. The document also highlights the use of Mojo IPC and rigorous automated testing required to synchronize state and maintain visual fidelity across this new multi-process boundary.
Embed this episode
No transcript yet.